‘Indispensability’ is the central concept underpinning the treatment of refusal to deal claims under EU competition law. Firms can normally refuse to share their infrastructure with would-be competitors, to supply an input, or to licence their intellectual property. Where the requested access is, however, deemed indispensable to effective competition in an adjacent market—an exceptional circumstance—dominant undertakings may find their default market freedom constrained, the rationale being that control of such an essential facility renders any refusal to deal disproportionately harmful.

However, the conventional wisdom that instances of refusal to deal constitute an abuse only in the presence of indispensability has been challenged from multiple directions. This article, available here, surveys the departures from the orthodoxy that can be found in the jurisprudence.

Section II introduces refusal to supply as an antitrust theory of harm.

It has long been acknowledged that Article 102 TFEU may, in certain instances, proscribe refusals to contract with rivals by dominant undertakings. Yet refusal to deal is a complex and contentious theory of harm, raising thorny questions from a competition policy perspective. The objections to refusal to deal can be grouped into three categories—relating to efficiency, equity, and the administrability of remedies.

First, an over-vigorous application of the refusal to deal principle may ultimately have counterproductive effects on overall levels of competition, leading to inefficient outcomes. Mandatory access facilitates greater competition in the short term, generating benefits in terms of lower prices, enhanced consumer choice, and so forth. In the longer term, however, an unduly trigger-happy approach to mandatory access may diminish a firm’s ex ante incentives to invest, innovate, or develop new infrastructure or products, for fear of subsequent expropriation of such efforts through competition enforcement. Moreover, would-be rivals have diminished incentives to develop their own competing facilities or to self-supply, if they can simply free ride on the efforts of their integrated competitor. Second, forcing an undertaking to share its property or other business resources, or to contract with rivals against its will, represents a significant incursion into the default freedom of action of economic operators. Lastly, competition agencies would have to take on the tasks of monitoring and enforcing compliance with the terms of any mandatory access requirement. Such a quasi-regulatory role is typically disfavoured by enforcers and may add an additional layer of institutional reluctance to addressing claims of refusal to deal through competition enforcement.

As a result, refusal to deal has not been applied expansively. The Court of Justice in the 1990s accepted that such conduct constitutes an abuse of dominance only in ‘exceptional circumstances’. Three conditions were identified to demonstrate such exceptionality: the existence of an indispensable input, the absence of any objective justification excusing the refusal, and the elimination of competition downstream. A further element is engaged where the refusal relates to access to intangible property protected by IP rights: that the refusal prevents the emergence of new product for which there is consumer demand. This approach is echoed in the Commission’s Guidance on its Enforcement Priorities for Article 102 TFEU, in which it identified three cumulative conditions, which mark any refusal as an ‘enforcement priority’: namely, indispensability, the elimination of effective competition, and consequent consumer harm.

Section III discusses the indispensability criterion.

Whereas anticompetitive foreclosure (requiring both exclusion of competitors and attendant consumer harm) has been adopted as the baseline standard for intervention across Article 102 TFEU, indispensability marks the point at which EU competition law shifts its focus from what undertakings cannot do to what they must. Indispensability thus functions as the lynchpin of refusal to deal as a theory of harm, providing a threshold requirement for intervention. Indispensability, self-evidently, requires more than dominance, although it typically coincides with a position of significant power in the relevant market. The concept, broadly speaking, is akin to that of an ‘essential facility’ namely, an obligatory input, access to which can be considered ‘critical to . . . competitive vitality’ in an adjacent market, and which cannot be secured from suppliers other than the defendant.

It is important, however, to distinguish between those circumstances where the dominant undertaking’s market position upstream merely gives it a ‘competitive advantage’ downstream, and those where it allows it ‘a genuine stranglehold on the related market’. The oversized negative spillover effects associated with blocking access in the latter instance justify antitrust scrutiny of how the facility owner uses its property. This leads, naturally, to the question of what precisely constitutes an ‘indispensable’ input. The crux of the concept is that access at the upstream level constitutes a ‘prerequisite for effective competition’ downstream. Indispensability, accordingly, entails both an absence of existing effective (albeit not necessarily equivalent) substitutes, and no reasonable prospect of self-supply.

Several aspects of this legal standard merit closer scrutiny. First, the test is neither one of practical convenience nor of absolute necessity. The CJEU adopted what is in essence a variation on the ‘as efficient competitor’ standard, requiring it to be demonstrated that duplication is objectively unfeasible by undertakings operating on an equivalent scale to the dominant firm. On the other hand, it is unnecessary to demonstrate that the refusal makes it entirely impossible to compete downstream. Thus, if all available substitutes are so unreasonable as to render effective competition unviable, even though some competition can strictly survive at the margins, then the threshold for indispensability has been reached. The question is thus whether a ‘viable’ or ‘realistic potential alternative’ exists. Barriers to the existence of such a viable alternative can be economic, legal (IP or otherwise) or even related to consumer behaviour.

Section IV considers four exceptions to the orthodoxy of indispensability.

In recent years, both courts and competition agencies adopted a variety of exceptions to the general rule described above. Some have been embraced explicitly, whereas others are implicit within nominally distinct theories of harm. This section examines four categories of such exceptions.

Regulated duties to deal – Article 102 TFEU is prepared to treat regulatory duties as proxies for indispensability, thus dispensing with the need to establish any free-standing objective necessity of access before examining the effects of the refusal.

Two distinct categories of regulatory obligations emerge from the case law. A first category concerns obligations imposed top-down by sector specific regulators. These were originally developed in the context of margin squeeze cases, and were then adopted by the Commission in its enforcement guidelines and, more recently in Slovak Telecom regarding mandatory obligations for local loop unbundling.

A second category concerns those obligations agreed by the dominant undertaking in the course of an essentially self-regulatory standard-setting exercise. This category, while generally applicable in the IP sphere, has arisen in more recent enforcement practice in the context of standard-essential patents (SEPs) within the mobile technology sector. Both the Commission and the Court have endorsed the view that ex ante commitments to licence agreed within the workings of a standard setting organization (SSO) might prevent a SEP holder from refusing to deal with willing licensees downstream.

Constructive refusals to deal – This denotes circumstances where the dominant firm nominally indicates willingness to contract with would-be customers, yet where the terms for access or the quality of supply delivered amount, in substance, to a refusal to do so. Despite raising similar concerns as formal refusals to deal, the law seems to have dispensed with the indispensability criterion as regards constructive refusals to deal. In effect, the case law now seems to insinuate that, when a refusal to supply is implicit rather than express, the requirements for it infringing competition law can be looser – despite it not being clear how this distinction should operate in practice, or why constructive and express refusals to supply should be treated differently. Indeed, it might even be reasoned that it is perverse to subject firms, which voluntarily contract with rivals (albeit on allegedly disadvantageous terms), to heightened antitrust scrutiny compared with those who refuse to do so outright.

This is particularly apparent as regards margin squeeze, which arises where dominant undertakings ‘charge a price for the product on the upstream market which, compared to the price it charges on the downstream market, does not allow even an equally efficient competitor to trade profitably in the downstream market on a lasting basis’. Every margin squeeze can be conceptualized as a constructive refusal to supply. However, following Telia Sonera, what matters for margin squeeze is the charging of dissimilar terms and conditions, not a refusal to grant access to the facility leading to foreclosure. Instead of indispensability, it is now sufficient to demonstrate the requisite ‘unfair spread’ between pricing levels, coupled with evidence of anticompetitive effects that may potentially exclude efficient competitors – despite there being no independent competitive harm caused by the margin squeeze above and beyond the harm which would result from a duty-to-deal violation at the wholesale level.

Flagrant refusals to deal – This category comprises a diverse group of cases involving the denial of access to an input to which downstream economic operators may be said to have some legitimate expectation, coupled with reprehensible or unconscionable behaviour by the dominant undertaking. The author provides four examples of this: Baltic Rail – a case involving the removal of 19 km of railway track by the Lithuanian rail incumbent to prevent a competitor downstream from providing a service -, AstraZeneca – a case involving several instances of regulatory gaming by a pharmaceutical company to extend the patented lifespan of its market-leading ulcer medicine, Losec, including withdrawing market authorisation to preclude market entry by generics and move patients to a different patent-protected medicine –, Rambus – a case concerning a patent ambush in the setting of a technological standard – and Sol Tec – a case concerning a refusal to fulfil customer orders in order to prevent parallel trading.  This is an amorphous but distinct category of refusals to supply that are prohibited by virtue of the sheer flagrancy of the defendant’s behaviour. This theory of harm hinges upon two requirements: (i) denial of access to an input or other resource to which downstream competitors may be said to have some legitimate expectation or reasonable entitlement (even if it is not, strictly speaking, indispensable to downstream competition); (ii) a moral component, requiring some form of reprehensible or unconscionable behaviour by the dominant undertaking.

Indispensability concerns only remedies – Some contend that the applicability of the Bronner criteria (indispensability) is determined by the nature of the remedy sought, rather than the characteristics of the underlying breach. In effect, this was an element of the Commission’s approach in its Google Shopping case, anchored in earlier case law such as Van der Bergh and the AG’s Opinion in AstraZeneca. The author objects to this approach on numerous grounds. First, it considers that it is not supported by precedent, which clearly distinguishes the existence of an infringement from the question of what remedies should apply. Second, this approach presupposes a clear dichotomy between the sorts of ‘mandatory’ remedies that trigger application of Bronner case law and purely prohibitive remedies outside its scope. Yet, an obligation to end a constructive refusal necessarily entails adapting one’s business behaviour to offer supply on fairer or more realistic terms, which unsustainably blurs the line between both types of remedies. Third, such an approach would create confusion regarding the legal standard applicable to business conduct. This concern is particularly acute in light of the continued decentralization of enforcement to NCAs and the development of private enforcement. Under this approach, a stand-alone private damages action for refusal to supply would not have to meet the indispensability criterion, which makes little sense.

Section V concludes by reflecting upon the status of the indispensability criterion today.

In most of the examples evaluated in the preceding section, the Commission and/or court have proceeded expressly by way of derogation, distinguishing the relevant circumstances from refusals to supply. Practically speaking, however, if each of the derogations analysed above is indeed applicable, then it is only in relation to outright refusals to supply, in markets where (quasi-) regulatory obligations have been neither imposed nor assumed, absent unconscionable conduct on the part of the defendant, and where the remedy sought is precisely a mandatory supply obligation that indispensability continues to constitute a basic requirement. If this is indeed the case, the refusal to supply doctrine is inoperative in practice, and easy to circumvent.

Indispensability today would appear to be something of an endangered species in EU competition law, whose habitats are continually being diminished and destroyed. Yet this is neither an inevitable nor a desirable development. Although the case law has twisted and turned in the past two decades, the underlying reasons for limiting refusal to supply cases to situations of indispensability to competition—and not mere convenience—are no less compelling. Indeed, with further decentralization of enforcement to NCAs and private actions brought before national courts, there is arguably an even greater case for limiting the applicability of the doctrine to objectively ‘exceptional circumstances.’


As one would expect from the author, this is a very comprehensive, thoughtful and rigorous paper. I particularly enjoy how the paper takes a well-known topic and turns it inside out – and persuasively argues that the existing exceptions to the ‘orthodox’ refusal to supply doctrine are such as to replace that doctrine in practice.

To my mind, the dynamic the author portrays is the result of a tension created by the very nature of the case law on refusal to supply. This is the theory of harm with the most stringent legal requirements. As such, defendants have incentives to frame sanctioned/investigated conducts as refusals to supply whenever possible. Given the nature of the abuse of dominant cases, which predominantly focus on exclusionary practices, many such conducts can indeed be conceptualised as refusals to supply. This leads agencies and courts continuously to distinguish investigated conducts from refusals to supply – in effect, competition agencies have incentives to avoid such a categorisation, and the stringent legal standards that go with it.

Given this, debates about whether various unilateral practices amount to refusals to supply should be expected to continue. In effect, one can even present examples from just the few months since this article was published. In September, AG Saugmandsgaard held, in Slovak Telekom, that the Bronner criteria should only apply to naked refusals to supply, and not to negotiated contractual terms that might be said to amount to implicit refusals to supply. More recently, in November, the General Court held in its Baltic Rail judgment that, if one were to assume that the infringement amounted to a refusal to supply (which the court deemed unnecessary), the Commission did not need to establish indispensability since the defendant was subject to a legal duty to provide access to the rail infrastructure. It is precisely due to this expansive nature of discussions about whether new unilateral practices amount to refusals to supply that one should be particularly careful in distinguishing between true exceptions to the refusal to supply doctrine, and practices with more spurious connections. It is here – and despite my appreciation for the author’s ability to devise insightful analytical categories – that I had some reservations. In particular, I think the categories adopted may be too broad. For example, most competition lawyers would be surprised to find that withdrawing a market authorisation, typically an element of product hopping, is in reality a form of refusal to supply. In effect, I had the impression that most cases falling within the ‘flagrant refusals to supply’ category would more naturally fit in other types of abusive practices. Likewise, I would have liked the distinction between ‘constructive refusals to supply’ and related practices to have been more clearly delineated, since this is likely to be the area most important to determining the effect scope of the ‘indispensability’ criterion.

Author Socials A weekly email with competition/antitrust updates. All opinions are mine

What do you think?

Note: Your email address will not be published